The location had still been well known in the mid-seventeenth century, and yet, despite being of such importance, memory of the battle became obscured over time. The site migrated across the landscape and, by the late eighteenth century, it was thought that the battle had been fought on Ambion Hill in Sutton Cheney. This was a belief securely fixed by Hutton in his influential book of 1788 and then confirmed by the work of subsequent authors.² It was therefore on Ambion that the Bosworth Battlefield Visitor Centre, the first of its kind in England, was established by Leicestershire County Council in 1976 (Figure 0.2).

Then in 1985, on the 500th anniversary of the battle, Colin Richmond pointed out that Ambion Hill seemed not to fit the details in the original accounts of the action.³ He suggested that the battlefield actually lay further to the south, in the adjacent township of Dadlington. For more than two decades there followed a sometimes vociferous debate as historians argued as to where the battle was really fought. By 2004 there were at least four serious possibilities, scattered across more than 6 km of the Leicestershire landscape, between Market Bosworth and Atherstone, each championed by different historians (Figure 0.3).⁴

Then, in 2003, plans were prepared by Leicestershire County Council (LCC) to revamp their interpretative facilities on Ambion Hill. They sought resources for this from the Heritage Lottery Fund (HLF) who then took advice from English Heritage. They argued that every effort should be made to ensure that the correct story of the battle was told in this major new exhibition and that visitors were directed to the true site of the action. Thus there was a decision to turn to battlefield archaeology in an attempt to settle the dispute that historians had proved unable to resolve.

The first step was the commissioning of a desk-based assessment, reviewing the major secondary works and the primary evidence, which was brought together with other relevant and easily available data from the Historic Environment Record and elsewhere.⁵ The data was then assessed in the context of battlefield studies in England and beyond. In this report we concluded that the Atherstone site promoted by Jones and Austin was highly improbable, and that the ‘traditional’ site on Ambion Hill, most recently championed by Williams, was also unlikely. The available evidence was found to point most strongly to the battlefield lying more than 1.5 km to the southwest of Ambion. This was the area suggested by Foss, who had elaborated Richmond’s hypothesis to provide the most convincing topographical context for the battle.⁶ However major uncertainties remained, even with the Foss interpretation. It was therefore recommended that a detailed field investigation be undertaken across five townships: the well-known candidates of Sutton Cheney, Dadlington, Shenton, Stoke Golding, together with the previously unconsidered township of Upton. The study was to employ the full methodology of battlefield archaeology which, over the previous decade, had begun to transform the study of historic battlefields in Europe.⁷

In 2005, with £154,000 funding from the HLF and additional support in kind from LCC, the Battlefields Trust was commissioned to undertake an investigation over three seasons to locate the battlefield.⁸ Work began on the Bosworth project in September 2005.

The approach we applied drew heavily upon investigations since 1984 by Scott and others on battlefields of the eighteenth and nineteenth centuries in the USA. But the methodology also inherits a great deal from a much longer tradition of research on battlefields in Britain by local historians, and from the interdisciplinary techniques applied over the last half century to the study of the English landscape. In this the Bosworth project forms part of a wider European initiative to use archaeological techniques to explore battlefields of an earlier period than those studied in the USA. It was clear, from the outset, however, that it would be necessary to push the boundaries of this evolving methodology if we were to resolve the Bosworth problem.⁹

Each step was systematically applied by a team comprising historians, palaeo-environmentalists, a soil scientist and a place name scholar, working alongside various archaeologists and a dedicated group of volunteers. The approach demands such a wide range of expertise because the evidence exists in so many forms. It can only be read effectively by integrating the approaches of military history with those of historical geography and landscape archaeology, underpinned by the systematic use of the metal-detector – the tool which lies at the heart of modern battlefield archaeology.¹⁰

The specific methodology used at Bosworth evolved from work on Civil War battlefields and siege sites in England, starting with Naseby in 1995.¹¹ The terrain based approach applied there lies in direct line of descent from the work of Burne, Newman and others in the twentieth century, although many of the strands can in turn be traced back to antiquarian work in the nineteenth century.¹² The Naseby study integrated analysis of primary sources for the military history with terrain reconstruction but for the first time it applied the techniques of landscape archaeology, an approach which was subsequently refined in work on other sites including Sedgemoor and Braddock Down.¹³ Systematic metal detecting survey was added in the late 1990s, initially with the assistance of Bob Kings and the MARS detecting group, in work on the siege site at Grafton Regis (1643).¹⁴ The approach was then substantially modified and refined to create a realistic method for battlefield-wide survey at Edgehill (1642) where, between 2004 and 2007, most aspects of the methodology were integrated in a comprehensive study.¹⁵

While there was now an effective approach for the investigation of Civil War sites, the investigation of medieval battlefields had proved far more problematic. This is because of the limitations of the contemporary written sources and the ephemeral nature of the battle archaeology.¹⁶ Important results had been achieved on the already well located site at Towton (1461), through a programme of geophysics, excavation and an intensive programme of non-systematic metal detecting. However, no medieval battlefield had yet been subject to large scale systematic investigation.¹⁷ The far more substantial level of funding available for the Bosworth survey would, for first time, allow the implementation of the whole methodology for battlefield investigation, using experts from all the relevant areas of expertise.

Such an approach is diametrically opposed to that taken by historians such as Jones, who stated in 2002 that:

trying to work out exactly what happened in a medieval battle is a redundant methodology … [I] take issue with Lieutenant-Colonel Alfred Burne’s theory of Inherent Military Probability, which still casts its baleful influence on Wars of the Roses battle reconstruction. I believe we need to explore the ritual employed before and during an engagement to find out more about why men fought … Our modern desire for a running order of events and a diagram of the action furnished with arrows, illustrating the positions and movement of the participants, is going to be frustrated … The order in which [the events] took place, and the cause and effect between them, is ultimately unknowable.’¹⁸

Such attitudes, also shared by a few archaeologists, push battlefield studies along a dangerous path away from the logical methods of modern science.¹⁹ The influence of these other factors was long ago and most eloquently presented by Keegan but, as a military historian, he had advanced this alongside a clear recognition of the need for a sound understanding of the ground and of the way the events played out within the landscape.²⁰

In our return to the approach of Burne, which we have elaborated through the application of modern techniques, we can draw clear justification from the military authors of the classical period onwards. Indeed the most influential of those works, by Vegetius, was actually owned and presumably read by military commanders of the Wars of the Roses, including Richard III, as it had been by their predecessors in earlier centuries. Of these classical authors none are clearer or more often repeated than Vegetius, who states that ‘a large part of a victory depends on the actual place in which the battle is fought … If we are strong in cavalry, we should opt for plains; if in infantry, we should choose confined places, obstructed by ditches, marshes or trees …’²¹

In the present study we also reject the notion of an independent archaeological study of battlefields of the historic period. It is in our view essential that, wherever possible, the documentary and the physical record of military events and of their context are explored by specialists in each of the relevant disciplines working in collaboration. As such, battlefield investigation represents an exemplar of the way in which the integration of the disciplines of history and archaeology, argued for by Smail, is in reality already being achieved. Indeed, in doing this we are merely following and extending the strong tradition established, since the late 1950s, by students of the medieval and early modern landscape of England.²²

The success of the recent work at Bosworth, however partial, which is detailed is the following chapters answers the claim of Griffiths and Thomas that, in resolving the problems of battlefield study, ‘there is no sound alternative to relying on the contemporary record’.²³ The different perspectives of historian and archaeologist on the same events will give rise to tensions, but if carefully exploited such tensions can prove a great strength. It ensures no assumptions are easily allowed to float through unchallenged, for each discipline will demand of the other carefully documented proof or demonstration of a lack thereof.

The methodological steps applied at Bosworth will be briefly rehearsed here as a context within which the following chapters can be read.²⁴ First one must to review the secondary sources to understand current interpretations and available evidence, and to locate the battlefield or the alternative sites that need to be examined. For most battles of the early modern period the general location of a battlefield has been established with some confidence. In contrast, for many, indeed perhaps the majority of medieval battles this is not the case, though the limitations in understanding will rarely have been highlighted in the stark way they had been at Bosworth. Many early battles simply await their Colin Richmond to highlight the problems.

Such uncertainty greatly complicates the investigation, for work has to be extended across a much wider landscape than would otherwise be the case. In this situation the objective of the first stage is to provide at least some focus for the initial investigation, by considering any alternative locations that have been proposed in the past. Then a rapid assessment is made of the main primary sources for the battle, any archaeological finds, local traditions and place name evidence that has already been collected. As we have seen, for Bosworth these first tasks were completed in the 2004 Assessment, which revealed four substantial candidate sites, each championed by a different author. These were prioritised and the conclusion was drawn that there was no justification for abandoning the previous consensus that that the battle had been fought somewhere in the townships immediately to the south-west of Sutton Cheney. Only if the research there failed to deliver the expected results, or new evidence was forthcoming from the Atherstone area, would further consideration be given to the hypothesis presented by Jones. In addition, it was concluded that the site on Ambion Hill was also unlikely to be correct, although this was something that had to be tested, as did the site suggested by Wright. However, it was the location championed by Foss which appeared the strongest contender (Figure 0.4).²⁵

The second task is to rework the primary accounts, where necessary transcribing and/or translating from the original to ensure accuracy, attempting to understand their exact meaning and to establish their biases and limitations. These data have various later uses, but the first is to identify, in full and accurate detail, the topographical clues as to where the action was fought. Even the names of a battle can provide important evidence in this regard that may not have been fully recognised or exploited. This is particularly true where the battle had several names, though few will have had so many as Bosworth. In addition to Redemore Field, by which Bosworth was known in the days immediately following the action, there were at least 11 other names or locational descriptions.

The third task is to characterise the armies and their equipping. This should ideally include information on numbers and deployment. While such evidence may be available in considerable detail for major battles of the early modern period, for the medieval period it rarely exists or, when it is does, it is often unreliable since the chronicle accounts are particularly prone to wild exaggeration. Where direct detail is lacking then reference can be made to more general practice of the period, with regard to numbers and especially to military practice, though even here there are major uncertainties compared with later centuries.²⁶ Such evidence is needed to allow the deployment of the armies to be reconstructed, or at least to some indication of the likely scale of the frontages to be established, for only then can the likely relationship of the armies to the terrain be assessed.

The fourth task is to reconstruct, as far as practicable, the landscape of the battlefield as it was at the time of the battle, based on documentary and physical evidence. This should include the reworking, from original sources, of any significant evidence previously analysed by others. The data is then integrated in digital form with other spatial data using a Geographical Information System (GIS). In this study MapInfo was used. In the case of Bosworth, for the reasons already explained, this aspect of the study needed to cover an exceptionally wide area, comprising some 28 km². Such work seeks to bring together all evidence on the historic landscape that may enable the topographical clues in the primary accounts to be unlocked. Of the greatest concern, at least in the later stages of the study, is the historic terrain of the battlefield, taken here to mean elements of the historic landscape which are likely to have had tactical significance.

The fifth task is to integrate the preceding evidence, placing the documented events within the historic terrain using the topographical clues in the primary accounts. This will be informed, where appropriate, by any relevant traditions, place names and archaeological evidence. If necessary, one may also resort to the broad principles of Burne’s ‘Inherent Military Probability’, but informed with reference to current understanding of military practice of the period. In this way it should be possible to develop one or more hypotheses as to where the battle was fought and, ideally, how the action played out within the landscape.

The sixth phase is normally one in which the interpretations arrived at in stage 5 are validated and refined by surveying the battle archaeology, which is an independent data set. The principal approach is a systematic metal-detecting survey, to recover a representative sample of the scatter of metal artefacts deposited during or immediately after the action and so reveal their distribution across the landscape.

At Towton there had never been any real doubt, at any time since 1461, as to the location of the battlefield. At Edgehill it has been possible to focus down onto two alternative scenarios for the events of 1642. Unfortunately at Bosworth, as is likely to be the case on many other medieval battlefields, the topographical clues in the documentary records are so poor that they do not allow the accurate placing of the action. Thus what we lacked throughout the project, until 2009, was a hypothesis that placed the events in a tight geographical framework. We were therefore thrown back onto sampling for battle archaeology in order to find the site. In the period before we located the battlefield Peter Liddle, then County Archaeologist in Leicestershire, likened it to the children’s game of ‘battleships’: dropping into individual fields scattered across the landscape and determining whether or not any battle archaeology was present. In the absence of positive evidence we then jumped to another square, and so on. This was never the intended approach and, as we were to learn much later, it was one likely to produce false negatives. This is because the limitations of time and money, relative to the vast area we needed to search, meant we were not able to survey at sufficient intensity. The latter is a crucial issue to which we will return at several points in the following chapters, since it has wide implications for the future study of medieval battlefields.

The survey area for stages 2–6 of the investigation, defined on the basis of the 2004 Assessment, encompassed the townships of Sutton Cheney, Dadlington, Shenton and Stoke Golding, together with Upton which had not previously been investigated by anyone in connection with the battle. As we have seen, this area encompassed all alternative hypotheses in the traditional zone. Subsequently, in 2006, the study area was expanded slightly south-westwards into the north-eastern corner of Higham on the Hill township although, as it turned out, further areas on the western periphery should also have been added.

Even though the Atherstone site had been dismissed as a result of the 2004 Assessment, John Austin was subsequently encouraged to undertake further research in that area, to see if any more specific evidence could be recovered in support of his hypothesis, which had been championed by Jones. Limited assistance, advice and training were given to the small local team which he brought together for the purpose. It was also agreed that, should the group reveal any substantial new data indicating that the battlefield lay near Atherstone, then we would consider investigation as part of the main project. Austin’s work was to include metal detecting by volunteers from the Hinckley detecting club but, in the absence of a clear target area, this was never implemented. Subsequently metal detecting was undertaken independently in the general area by Andrew Tansley, in consultation with LCC. Unfortunately this work did not employ a systematic methodology, so we have not been able to integrate the results, as comparative data, in our analysis. As Austin’s work failed to yielded significant new information, so the Atherstone area was never included in the Bosworth project.

Clearly there were several reasons why the Bosworth project was to prove such an enormous challenge. As a result it was not practicable to complete the study within the three years as originally planned – especially as the metal detecting survey depended on volunteers and thus was generally limited to a single day in any week. Despite this, definitive evidence for the battle was eventually recovered, though in good archaeological tradition it was not until the final week of a 6 month extension to the project!

On 1 March 2009 a single lead ball of just 30 mm diameter was recovered in the last unsurveyed sector in the western periphery of the study area. It could be nothing other than a round shot fired during the 1485 battle. A week later a second and much larger round shot was recovered in the adjacent field to the west. This was enough to justify an extra month of intensive fieldwork. The results achieved in those 4 weeks were such that a full year’s extension was approved, to enable the extent and character of the battlefield to be defined. In fact fieldwork did not finally end until late 2010. By then sufficient archaeological evidence had been recovered to compare the patterns with the evidence in the primary accounts and to place both within the reconstructed historic terrain. The three largely independent data sets were found to correspond in a particularly meaningful way.